xfs: don't check for AG deadlock for realtime files in bunmapi
authorOmar Sandoval <osandov@fb.com>
Wed, 27 Nov 2019 00:58:08 +0000 (16:58 -0800)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Sun, 12 Jan 2020 10:24:16 +0000 (11:24 +0100)
commita76976082656b9b74db72cdeb35cd114ecaadcd0
treee98e3b0e4cee6b9a3d395b247d3e6a41b3619964
parentaf49ad548b93659e243dedecf1432b0f5eba3844
xfs: don't check for AG deadlock for realtime files in bunmapi

commit 69ffe5960df16938bccfe1b65382af0b3de51265 upstream.

Commit 5b094d6dac04 ("xfs: fix multi-AG deadlock in xfs_bunmapi") added
a check in __xfs_bunmapi() to stop early if we would touch multiple AGs
in the wrong order. However, this check isn't applicable for realtime
files. In most cases, it just makes us do unnecessary commits. However,
without the fix from the previous commit ("xfs: fix realtime file data
space leak"), if the last and second-to-last extents also happen to have
different "AG numbers", then the break actually causes __xfs_bunmapi()
to return without making any progress, which sends
xfs_itruncate_extents_flags() into an infinite loop.

Fixes: 5b094d6dac04 ("xfs: fix multi-AG deadlock in xfs_bunmapi")
Signed-off-by: Omar Sandoval <osandov@fb.com>
Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
fs/xfs/libxfs/xfs_bmap.c